Large scale phylogenomic analysis resolves. a backbone phylogeny in ferns

Large scale phylogenomic analysis resolves a backbone phylogeny in ferns Hui Shen 1,2#, Dongmei Jin 1,2#, Jiang-Ping Shu 1,2, Xi-Le Zhou 1,2, Ming Lei 3, Ran Wei 4, Hui Shang 1,2, Hong-Jin Wei 1,2, Rui Zhang 1,2, Li Liu 1,2, Yu-Feng Gu 1,2, Xian-Chun Zhang 4, Yue-Hong Yan 1,2* # Equal contributors *Corresponding author: yhyan@sibs.ac.cn 1 Shanghai Chenshan Plant Science Research Center, Chinese Academy of Sciences, Shanghai 201602, China 2 Shanghai Key Laboratory of Plant Functional Genomics and Resources, Shanghai Chenshan Botanical Garden, Shanghai 201602, China 3 Majorbio Bioinformatics Research Institute, Shanghai 201320, China 4 State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China. Abstract

Background: Ferns, originated about 360 million years ago, are the sister group of seed plants. Despite the remarkable progress in our understanding of fern phylogeny, with conflicting molecular evidence and different morphological interpretations, relationships among major fern lineages remain controversial. Results: With the aim to obtain a robust fern phylogeny, we carried out a large scale phylogenomic analysis using high-quality transcriptome sequencing data, which covered 69 fern species from 38 families and 11 orders. Both coalescent-based and concatenation-based methods were applied to both nucleotide and amino acid sequences in species tree estimation. The resulting topologies are largely congruent with each other, except the placement of Angiopteris fokiensis, Cheiropleuria bicuspis, Diplaziopsis brunoniana, Matteuccia struthiopteris, Elaphoglossum mcclurei and Tectaria subpedata. Conclusions: Our result confirmed that Equisetales is sister to the rest of ferns, and Dennstaedtiaceae is sister to eupolypods. Moreover, our result strongly supported some relationships different from the current view of fern phylogeny, including that Marattiaceae may be sister to the monophyletic clade of Psilotaceae and Ophioglossaceae; Gleicheniaceae and Hymenophyllaceae form a monophyletic clade sister to Dipteridaceae; and that Aspleniaceae is sister to the rest groups in eupolypods II. These results were interpreted with morphological traits, especially sporangia characters, and a new evolutionary route of sporangial annulus in ferns was suggested.

This backbone phylogeny in ferns sets a foundation for further studies in biology and evolution in ferns, and therefore in plants. Key Words: phylogenomic, monilophytes, evolution, sporangium, transcriptome Background Phylogeny, which reflects natural history, is fundamental to understanding evolution and biodiversity. Ferns (monilophytes), originated about 360 million years (MY) ago, are the sister group of seed plants [1, 2]. With estimated 10,578 extant living species globally [3], they are the second most diverse group of vascular plants. Phylogenetic studies for ferns, especially based on molecular evidence, have been widely carried out in recent decades. These studies have revolutionized our understanding of the evolutionary history of ferns. Milestones included setting ferns as the sister group of seed plants [1, 2], placing Psilotaceae and Equisetaceae within ferns [2, 4, 5], and revealing a major polypods radiation following the rise of angiosperms [6, 7]. Resolutions at shallow phylogenetic depth among families or genera have also been improved remarkably [8-14]. However, previous research on fern phylogeny has mostly relied on plastid genes [10, 12, 13], some combined with a few nuclear genes [4, 5, 14] or morphological traits [5, 11]. Due to incomplete lineage sorting (ILS), genes from different resources often show conflicting evolutionary patterns, especially when based on a limited number of

samples, some deep relationships in fern phylogeny remain controversial (Figure 1). In the latest PPG I system [3], which has derived from many recent phylogenetic studies, some important nodes remain uncertain, such as (i) what are the relationships among Marattiales, Ophioglossales and Psilotales? (ii) are Hymenophyllales and Gleicheniales sister groups? and (iii) what are the relationships among families in eupolypods II? Transcriptome sequencing (RNA-Seq) provides massive transcript information from the genome. Phylogenetic reconstructions based on RNA-Seq are more efficient and cost-effective than traditional PCR-based or EST-based methods when lacking whole-genome data [15]. Successful cases in recent years include mollusks [16], insects [17], the grape family [18], angiosperms [19], and land plants including six ferns [20]. Here, with the aim to reconstruct the framework of fern phylogeny, we sampled abundant fern species representing all important linages and applied latest phylogenomic analyses based on RNA-Seq. To reconstruct a robust and well-resolved phylogeny in ferns, applying multiple methods of phylogenomic analysis is extremely important. Since concatenation-based estimations of species trees usually have good accuracy under low level of ILS, while coalescent-based methods are developed to overcome the effect of ILS, but are sensitive to gene tree estimation error [21], both concatenation-based and coalescent-based estimations are applied. Nucleotide sequence, with higher variability than amino acid sequence, usually brings more useful information in phylogeny

reconstruction, especially for closely related taxa. However, the substitutional saturation and compositional bias in nucleotide sequence, especially in the third codon position, may lead to a deviation from the true phylogeny. Here, both nucleotide and amino acid sequences are used in phylogeny reconstruction. Morphologically, the fern sporangium is an organ for enclosing and dispersing spores, most of which functions like a unique catapult with the annulus [22]. During the last centuries, Bower s hypothesis on the evolution of sporangia with a focus on annulus [23] had been one of the most important cornerstones to fern phylogeny based on morphology [24, 25]. However, this hypothesis has been challenged by somewhat conflicting frameworks of fern phylogeny [4, 10, 12, 14, 26]. A robust framework in fern phylogeny which reflects the evolutionary history will improve our understanding for the evolution of fern sporangia as well as other characters. Data description Taxa sampling and RNA-Seq We chose 69 fern species from 38 families according to PPG I system (totally 48 fern families), covering all the 11 orders (Equisetales, Psilotales, Ophioglossales, Marattiales, Osmundales, Hymenophyllales, Gleicheniales, Schizaeales, Salviniales, Cyatheales, and Polypodiales). Information about the location and time for sampling is given in Table S1. All the sampled species were collected under the permissions of the natural reserves and Shanghai Chenshan Botanical Garden in China.

Sporophyll or/and trophophyll were collected and frozen in liquid nitrogen immediately, and preserved in Ultra-low temperature refrigerator at -80 C before RNA extraction. Total RNA was extracted using TRIzol (Life Technologies Corp.) according to the manufacturer s protocols. The RNA concentration was determined using a NanoDrop spectrophotometer, and RNA quality was assessed with an Agilent Bioanalyzer. Paired-end reads were generated by Majorbio Company (Shanghai, China) using the HiSeq 2500 system. Raw reads were deposited in NCBI [27]. Transcriptomes assembly and orthology assignment Transcriptomes data were generated from 69 fern species (Table 1). After filtering, about 2,726.9 million pair-end DNA sequence reads (about 313 Gbp) were retained. We assembled these reads de novo and obtained a total of 5,449,842 contigs [28]. In order to obtain a reliable phylogenetic relationship, we selected four species as the outgroup, representing the main lineages of land plants: Amborella trichopoda (representing angiosperms), Picea abies (representing gymnosperms), Selaginella moellendorffii (representing lycophytes) and Physcomitrella patens (representing bryophytes). The translated ORF (protein) sequences of these four species were downloaded from Phytozome [29] and used in the following analysis. To ensure the consistency of phylogenomic analysis, we used a phylogenetic-based ortholog selection method, and obtained two subsets of one-to-one orthologous genes that differed in gene number and species occupancy rate, named Matrix 1 and Matrix 2 [30]. Matrix 1 consists of 2391 genes that are present in at least 52 taxa (that is 75% of the 69 taxa in total), resulted in 2,024,565 nucleotide and 674,855 amino acid positions, the gene and character occupancy were 88% and 85%

respectively. Matrix 2 consists of 1334 genes that are present in at least 62 taxa (that is 90% of the 69 taxa in total), resulted in 1,171,332 nucleotide and 390,444 amino acid positions, the gene and character occupancy reached 94% and 90% respectively. For each orthologues gene set, coalescent-based and concatenation-based methods were applied separately to both nucleotide and amino acid sequences. A working flow diagram showing the major processes in this study is presented in Figure 2. Results Species tree estimated in 69 ferns For each combination of reconstruction methods (coalescent-based or concatenation-based) and sequence types (nucleotide or amino acid), Matrix 1 and Matrix 2 [31, 32] always yielded the same topology. In general, the four topologies (Figure 3, Figure S1, S2, S3) from a combination of methods and sequence types are consistent except six positions (Table 2). Among the topologies, the one estimated by applying coalescent-based method to nucleotide sequence (Figure 3) and the one applying concatenation-based method (Figure S2) are most congruent. Reconstruction of the evolutionary history of sporangial annulus Our reconstruction of the evolution of sporangial annulus (Figure 4) showed that ex-annulus sporangia are inferred to be the ancestral state (proportional likelihood [PL]:

1), and the rest of annulus states are likely derived from ex-annulus sporangia. Vertical annulus is suggested as synapomorphy for all polypod ferns (PL > 0.99). Both oblique annulus and rudimentary annulus have experienced parallel evolution. Discussion Comparison of topologies estimated by various methods By comparing topologies estimated by coalescent-based and concatenation-based method using both nucleotide and amino acid sequences (Table 2), we found that the topologies yielded from coalescent-based and concatenation-based methods using nucleotide sequence are mostly consistent, except the position of Angiopteris fokiensis. Topologies yielded from coalescent-based method using nucleotide sequence and amino acid sequence showed three positions of inconsistency, all of which belong to eupolypods. Since eupolypods have experienced rapid evolutionary radiation in Cenozoic [7], and nucleotide sequences usually provide more information to reconstruct relationships at shallow phylogenetic scale, we consider the topology yielded from nucleotide sequence maybe more reliable. However, the inconsistent positions among topologies often show relatively lower supporting values, and are often the controversial nodes from past studies based on different genes, we suggest such inconsistency might be caused partially by LIS and reticulate evolution. Relationships of eusporangiate ferns Which clade is sister to the remaining taxa in ferns is a long-debated question (Figure 1). Our results strongly supported that Equisetales (horsetails) are the sister group to all other monilophytes. This topology confirmed the results reported by Rai & Graham [12],

and Kuo et al. [33] based on plastid genes, and has been accepted by the PPG I [3] in 2016. Distinct from most fern phylogeny based on molecular evidence (Figure 1), our results based on coalescent method revealed that Psilotales (whisk ferns), Ophioglossales (moonworts), and Marattiales (king ferns) form a monophyletic clade as ((Psilotales, Ophioglossales), Marattiales), which is sister to leptosporangiate ferns. The monophyletic origin of Psilotales, Ophioglossales, and Marattiales, which belong to eusporangiate ferns, is supported by the structure of sporangia. Being different from the leptosporangiate type, sporangia of eusporangiate ferns have no sporangiophore, they are thick in wall and large in volume, produce a large amounts of spores, and have no sporangial annulus or only have a few enlarged parenchyma cells. The incongruence between the results based on coalescent and concatenation methods may be caused by strong ILS effect, which is a main pitfall when using the concatenation method [21]. Relationship of early leptosporangiates Within early leptosporangiates, our results revealed a new monophyletic clade that Gleicheniaceae (forking ferns) is sister to Hymenophyllaceae (filmy ferns), which is different from the mainstream [3, 10, 12-14, 34]. Similar but still different from the topology (((Dipteridaceae, Matoniaceae), Gleicheniaceae), Hymenophyllaceae) reported by Pryer et al. in 2004 [5], in our results, Cheiropleuria, which belongs to Dipteridaceae and formerly placed in Gleicheniales [2, 5, 12, 26, 35, 36], is sister to the monophyletic clade of (Gleicheniaceae, Hymenophyllaceae).

This new relationship is supported by sporangia character. Early leptosporangiates [36] are characterized with diverse sporangia and annulus. However, both Gleicheniaceae and Hymenophyllaceae have spherical sporangia with transverse-oblique annulus, as well as a short sporangial stalk connecting to a prominent receptacle [37]. On the other hand, flattened sporangia with slightly oblique annulus are found in Cheiropleuria. Moreover, long sporangial stalk and inapparent receptacle are common in Cheiropleuria, Dipteris and Matonia. We suggest Dipteridaceae, probably together with its sister lineage Matoniaceae [5, 12], may be sister to the clade of (Gleicheniaceae, Hymenophyllaceae). According to our results, Gleicheniales, which is comprised of Dipteridaceae, Matoniaceae, and Gleicheniaceae [26], is no longer a monophyletic lineage, but a paraphyletic one. Relationships within polypod ferns Polypods include more than 80% of living ferns, and their phylogeny remains somewhat controversial and elusive [26, 35, 36]. Our results strongly supported that Dennstaedtiaceae instead of Pteridaceae is sister to eupolypods. This pattern confirmed the topology suggested recently by Rothfels et. al basing on 25 low-copy nuclear genes [14] and Lu et. al based on plastid genes [13], as well as PPG I system [3]. According to our results, relationships of Pteridaceae [34, 36, 38] and Dennstaedtiaceae [36] are also well resolved. Notably, Monachosorum is sister to the

rest members in Dennstaedtiaceae, rather than being sister to the lineage of Pteridium, Hypolepis and Histiopteris [36]. Our results showed that eupolypods are divided into two major lineages, eupolypods I and eupolypods II in agreement with the consensus opinion [3]. Within eupolypods II, our results supported that Aspleniaceae is the sister group to the rest members, which is different from the current viewpoints [26, 36, 39]. Within eupolypods I, our result strongly supported that Lomariopsidaceae and Nephrolepidaceae form a paraphyletic group, rather than a monophyletic clade based on plastid genes [10, 26, 36]. Our new topology confirmed the morphology-based hypothesis that Dennstaedtiaceae with two indusial, rather than Pteridaceae with one false indusium, is more closely related to eupolypod ferns [40]. In Pteridaceae, the unstable structure of spherical sporangia, including variable annulus and short sporangial stalk, indicates these characters of sporangia are relatively original and are close to those with oblique annulus in early leptosporangiates [23]. We also noticed that the characters of spherical sporangia with slightly oblique annulus in Monachosorum should be more ancestral than the flattened sporangia with typical vertical annulus in other genera of Dennstaedtiaceae. For distinguishing eupolypods I and eupolypods II, the number and shape of the vascular bundles at the base of petiole have been demonstrated to be a powerful diagnostic character [36, 39].

The evolution of sporangial annulus in ferns By observing the character of sporangial annulus of abundant samples in each fern group, and combining these characters with our well-resolved backbone phylogeny (Figure 3), we reconstructed the evolutionary history of sporangial annulus in ferns (Figure 4). According to the results, we infer that ex-annulus sporangia, as in Equisetaceae, Psilotaceae, and Ophioglossaceae, is the ancestral state in ferns; rudimentary multiseriate annulus, which is inverse U-shaped in Marattiaceae, and U-shaped in Osmundaceae; equatorial transverse-oblique uniseriate annulus, as in Gleicheniaceae and Hymenophyllaceae; oblique annulus as in Cyatheales (tree ferns), and vertical annulus as synapomorphy in polypods, have been derived from the ex-annulus state. Both apical annulus as in Lygodium and Schizaea, and vestige or disappeared annulus as in Salviniales (aquatic ferns) are likely to be specialized in parallel from oblique annulus. Inconsistent with Bower s hypothesis [23], our results showed that sporangia with apical annulus as in Schizaeales are no longer the ancestral type in ferns but a specialized one. Correspondingly, the oldest fossils of Schizaeaceae is now believed to appear in Jurassic (201-145 MY BP) rather than formerly thought Carboniferous (359-252 MY BP) [41]. Conclusion Our results confirmed that Equisetales is sister to all the other monilophytes, and Dennstaedtiaceae is sister to eupolypods which have been reported previously.

Moreover, our results revealed some new relationships, such as eusporangiate ferns except Equisetales may form a monophyletic clade as ((Psilotaceae, Ophioglossaceae), Marattiaceae); while Gleicheniaceae and Hymenophyllaceae form a monophyletic clade which is sister to Dipteridaceae; and Aspleniaceae is sister to the rest groups in eupolypods II. Most of these results are supported by sporangia characters, and a new evolutionary route of sporangial annulus in ferns is suggested. Potential implications Here, we present a robust fern phylogeny yielded from a largescale phylogenomic analysis based on a high-quality RNA-seq dataset covering 69 fern specie. This backbone phylogeny in ferns sets a foundation for further studies in biology and evolution in ferns and therefore in plants, especially when fern genomes are not available. Methods De novo transcriptome assembly For each paired-end library, we first removed the Illumina adapter of raw reads using Scythe (Scythe, RRID:SCR_011844) [42] and trimmed the poor quality bases using DynamicTrim Perl script of the SolexQA package with default parameters [43]. Next, de novo transcriptome assembly of each species was conducted using the Trinity

package, version: trinityrnaseq_r20140413 (Trinity, RRID:SCR_013048) with default parameters [44]. To discard the duplicated sequences, the obtained contigs were clustered using CD-HIT-EST v4.6.1 (CD-HIT, RRID:SCR_007105) to generate non-redundant contigs. All contigs longer than 200 bp in length were used for downstream analysis. We used TransDescoder, a program in the Trinity package, to identify the candidate coding sequences (CDSs) from the contigs with default criteria. Finally, the translated protein sequences of CDSs were searched by BLASTP against the non-redundant protein database in NCBI with an e-value threshold of 1e-5. These BLASTP hit sequences were used for further analysis. Orthology assignment, alignment, and alignment masking For orthology assignment for the 69 sample assemblies together with the four outgroup species a phylogenetic based clustering method described previously [16] was used. In short, all-vs-all BLAST search of amino acid sequence was performed across different species, the BLAST results were clustered using MCL [45] software with the parameters -I 2 tf gq(20). Optimization of the inflation parameter (I) was conducted as described previously [46], the default value 2.0 was selected ultimately. As the de novo assembly by Trinity produces many sequences with high similarity, which contain both paralogs and isoforms [47], when a clustered gene family contains too many sequences (eg. more than 10), the risk of contamination of isoforms rises, along with the computational infeasibility. Hence, when a species had more than 10 sequences in a gene family, we removed all sequences in this gene family of this species. Then, groups with at least 35 (50%) fern species were aligned using einsi command, implemented in MAFFT (MAFFT, RRID:SCR_011811) [48], and trimmed by Gblocks with default parameters [49]. Next, for each group, homologous gene tree was built with RAxML software, version: 8.0.20 (RAxML, RRID:SCR_006086), by implementing the maximum

likelihood method (ML) [50]. To infer orthologous genes, we used treeprune in the Agalma package [51] to mask the monophyletic sequences. We pruned the paralogous subtrees from the homologous gene trees until only one monophyletic subtree retained. Next, the resulted orthologous gene trees were further filtered by the criteria that each species should be represented by only one sequence, this resulted subset genes were referred to one-to-one orthologs, which were largely free of gene duplication. Then, we extracted both the CDSs (nucleotide sequence) and translated amino acid sequence from each orthologous gene group, followed by aligning with MAFFT and trimming with Gblocks. The alignment with coding and corresponding translated sequences longer than 150 bp (or 50 amino acids) in length were kept for the further analysis. BUSCO analysis The Basic Universal Single Copy Orthologs (BUSCO, RRID:SCR_015008), which employs a core set of orthologs conservative in eukaryotic species to determine the gene coverage of each assembly [52], was employed to assess the completeness of the transcriptome assembly we obtained (Table S2) [53]. A total of 303 BUSCOs were employed to blast against by translated amino acid of the assemblies using BLASTP. Then the number of complete and partially matched gene from each assembly was counted respectively. Out of the 69 samples in total, the gene coverage of 65 samples (94.2%) exceeded 82%, with at least 251 complete genes identified. Unexpectedly, among our total assemblies, 1 sample (Aleuritopteris chrysophylla, named RS_72) presented extremely low gene coverage degree, in which only 72 (23.8%) complete housekeeping genes were found (Supplementary Table 2). However, when the sample was deleted from the matrix used to construct the backbone of the phylogenetic tree,

the topology remained unchanged, indicating that the lower completeness in this sample did not affect our results (data not shown). Phylogenetic analysis The coalescent-based species trees were reconstructed by ASTRAL v4.10.4 [54], carried out 100 replicates of multi-locus bootstrapping [55]. Each gene tree was constructed with the PROTCATJTTF model by RAxML v8.2.4 (RAxML, RRID:SCR_006086) [50], performed 100 random replicates to calculate bootstrap value. For the concatenation analysis, we preformed the maximum likelihood analyses (ML) for each matrix using RAxML software (version: 8.0.20). The branch support was evaluated using 100 bootstrap replicates. We used the GTR + Γ4 + I model for DNA matrices, and the JTTF model for the corresponding protein matrices, selected by ProtienModelselection.pl [56]. To estimate the divergence times, we used the concatenated alignment of orthologs, calibrated with ages of two fossils (Archaeocalamites Senftenbergia: 354 MY, Grammatopteris: 280 MY [6, 57]) as the minimum ages of monilophytes and leptosporangiate ferns, respectively, and a maximum age constraint of 500 MY for land plants, in a Bayesian relaxed clock method using MCMCTREE [58] on the coalescent-based species tree. Reconstruction of the evolution of sporangial annulus Characters of sporangial annulus of the sampled species were observed using a polarized light microscope (Axio Scope.A1, ZEISS) after the fresh and mature sporangia were treated with sodium hypochlorite (NaClO) solution. The evolution of sporangial annulus was reconstructed with likelihood method implemented in Mesquite v2.7.5 [59]. All character states (i.e., vertical annulus, oblique annulus, rudimentary

annulus, ex-annulus, apical annulus, transverse annulus, and vestigial annulus) were treated as unordered and equally weighted. To reconstruct character evolution, a maximum likelihood approach using Markov k-state 1 parameter model [60] was applied. To account for phylogenetic uncertainty, the Trace-characters-over-trees command was used to calculate the ancestral states at each node including probabilities in the context of likelihood reconstructions. To carry out these analyses, characters were plotted onto 100 trees that were sampled in the ML analyses of the combined dataset using RAxML v7. The results were finally summarized as percentage of changes of character states on a given branch among all 100 trees utilizing the option of Average-frequencies-across-trees. Declarations List of abbreviations BUSCOs, the basic universal single copy orthologs; ILS, incomplete lineage sorting; MY, million years; PPG, the pteridophyte phylogeny group; RNA-Seq, transcriptome sequencing. Additional files

Additional file1: Tables S1 to S2 and Figures S1 to S3. Availability of data and materials Raw reads of RNA-Seq for 69 fern species were deposited in GenBank under Bioproject accession number PRJNA281136. Transcriptome datasets, alignments, phylogenetic trees, BUSCO results and other supporting data are available via the GigaScience repository GigaDB [61]. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Funding This work was funded by Shanghai Landscaping & City Appearance Administrative Bureau of China, Scientific Research Grants (G142433, G152420 and F112422) and the National Natural Science Foundation of China (31370234). Authors contributions YHY, HShen and DMJ conceived and designed the study. ML, JPS, DMJ, RW and LL implemented the data analyses. YHY, HShen, HJW, XLZ, HShang and YFG collected the specimens. HShen, RZ and YFG prepared the specimens for sequencing. XLZ provides the anatomical data. DMJ, HShen, YHY, JPS, ML, RW, HShang, XLZ and XCZ interpreted the results and wrote the manuscript.

Acknowledgements We thank Prof. Yong-Hong Hu, Prof. Jin-Shuang Ma, Prof. Zhao-Qing Chu and Dr. Jun Yang from Shanghai Chenshan Botanical Garden of China, as well as Prof. Fu-Wu Xing from South China Botanical Garden of CAS for helpful comments and suggestions. We thank Prof. Paul G. Wolf from Utah State University, Prof. Yin-Long Qiu from University of Michigan and Dr. Jin-Long Zhang from Kadoorie Farm and Botanic Garden for providing important suggestions regarding the research method. We appreciate helpful comments and suggestions from three reviewers of previous versions of this manuscript. Ethics approval and consent to participate Not applicable. References 1. Duff RJ, Nickrent DL. Phylogenetic relationships of land plants using mitochondrial small-subunit rdna sequences. Am J Bot, 1999;86:372-86. 2. Pryer KM, Schneider H, Smith AR, Cranfill R, Wolf PG, Hunt JS, et al. Horsetails and ferns are a monophyletic group and the closest living relatives to seed plants. Nature. 2001;409:618-22. 3. The Pteridophyte Phylogeny Group. A community-derived classification for extant lycophytes and ferns. J Syst Evol. 2016. doi:10.1111/jse.12229 4. Qiu Y-L, Li L, Wang B, Chen Z, Knoop V, Groth-Malonek M, et al. The deepest divergences in land plants inferred from phylogenomic evidence. Proc Natl Acad Sci USA. 2006;103:15511-6. 5. Pryer KM, Schuettpelz E, Wolf PG, Schneider H, Smith AR, Cranfill R. Phylogeny and evolution of ferns (monilophytes) with a focus on the early leptosporangiate divergences. Am J Bot. 2004;91:1582-98.

6. Schneider H, Schuettpelz E, Pryer KM, Cranfill R, Magallon S, Lupia R. Ferns diversified in the shadow of angiosperms. Nature. 2004;428:553-7. 7. Schuettpelz E, Pryer KM. Evidence for a Cenozoic radiation of ferns in an angiosperm-dominated canopy. Proc Natl Acad Sci USA. 2009;106:11200-5. 8. Zhang LB, Zhang L, Dong SY, Sessa EB, Gao XF, Ebihara A. Molecular circumscription and major evolutionary lineages of the fern genus Dryopteris (Dryopteridaceae). BMC Evol Biol. 2012;12:180-94 9. Liu H-M, Zhang X-C, Wang W, Qiu Y-L, Chen Z-D. Molecular phylogeny of the fern family dryopteridaceae inferred from chloroplast rbcl and atpb genes. Int J Plant Sci. 2007;168:1311-23. 10. Liu H-M. Embracing the pteridophyte classification of Ren-Chang Ching using a generic phylogeny of Chinese ferns and lycophytes. J Syst Evol. 2016;54:307-35. 11. Schneider H, Smith AR, Pryer KM. Is morphology really at odds with molecules in estimating fern phylogeny? Syst Bot. 2009;34:455-75. 12. Rai HS, Graham SW. Utility of a large, multigene plastid data set in inferring higher-order relationships in ferns and relatives (monilophytes). Am J Bot. 2010;97:1444-56. 13. Lu J-M, Zhang N, Du X-Y, Wen J, Li D-Z. Chloroplast phylogenomics resolves key relationships in ferns. J Syst Evol. 2015;53:448-57. 14. Rothfels CJ, Li F-W, Sigel EM, Huiet L, Larsson A, Burge DO, et al. The evolutionary history of ferns inferred from 25 low-copy nuclear genes. Am J Bot. 2015;102:1089-107. 15. Hittinger CT, Johnston M, Tossberg JT, Rokas A. Leveraging skewed transcript abundance by RNA-Seq to increase the genomic depth of the tree of life. Proc Natl Acad Sci USA. 2010;107:1476-81. 16. Smith S, Wilson N, Goetz F, Feehery C, Andrade S, Rouse G, et al. Resolving the evolutionary relationships of molluscs with phylogenomic tools. Nature. 2011;480:364-7. 17. Misof B, Liu S, Meusemann K, Peters RS, Donath A, Mayer C, et al. Phylogenomics resolves the timing and pattern of insect evolution. Science. 2014;346:763-7.

18. Wen J, Xiong Z, Nie Z-L, Mao L, Zhu Y, Kan X-Z, et al. Transcriptome sequences resolve deep relationships of the grape family. Plos One. 2013;8:e74394. 19. Zeng L, Zhang Q, Sun R, Kong H, Zhang N, Ma H. Resolution of deep angiosperm phylogeny using conserved nuclear genes and estimates of early divergence times. Nature Communications. 2014. doi:10.1038/ncomms5956 20. Wickett NJ, Mirarab S, Nam N, Warnow T, Carpenter E, Matasci N, et al. Phylotranscriptomic analysis of the origin and early diversification of land plants. Proc Natl Acad Sci USA. 2014;111:E4859-68. 21. Mirarab S, Bayzid MS, Boussau B, Warnow T. Statistical binning enables an accurate coalescent-based estimation of the avian tree. Science, 2014.doi:10.1126/science.1250463 22. Noblin X, Rojas N, Westbrook J, Llorens C, Argentina M, Dumais J. The fern sporangium: a unique catapult. Science. 2012;335:1322. 23. Bower FO. The Ferns (Filicales) treated comparatively with a view to their natural classification. Vol 1-3. London: Cambridge University Press. 1923-1928. 24. Pichi-Sermolli REG. Historical review of the higher classification of the Filicopsida. In: Jermy AC, Crabb JA, Thomas BA, editors. Phylogeny and classification of the ferns. London: Bot J Linn Soc. 1973. p.11-40. 25. Smith AR. Non-molecular phylogenetic hypotheses for ferns. Am Fern J. 1995;85(4):104-22. 26. Smith AR, Pryer KM, Schuettpelz E, Korall P, Schneider H, Wolf PG. A classification for extant ferns. Taxon. 2006;55:705-31. 27. Raw reads. https://www.ncbi.nlm.nih.gov/bioproject/?term=prjna281136. Accessed 5 July 2017. 28. Transcriptome datasets. https://figshare.com/s/0f773861b6813f97ff63. Acessed 5 July 2017. 29. Phytozome. http://phytozome.jgi.doe.gov/. Accessed 5 July 2017.

30. Alignments. Available from: https://figshare.com/s/f835735cb66911ff1ffd. Accessed 5 July 2017. 31. Datasets of coalescent-based species tree. Available from: https://figshare.com/s/e5e70c2fd3990e5176d8. Accessed 5 July 2017. 32. Datasets of concatenation based phylogenetic tree. Available from: https://figshare.com/s/8af236b660f61078e40b. Accessed 5 July 2017. 33 Kuo LY, Li FW, Chiu WL, Wang CN. First insights into fern matk phylogeny. Mol Phylogenet Evol. 2011; 59: 556-66. 34. Schneider H. Evolutionary morphology of ferns (monilophytes). Annu Plant Rev. 2013;45:115-40. 35. Christenhusz MJM, Chase M. Trends and concepts in fern classification. Ann Bot. 2014;113:571-94. 36. Schuettpelz E, Pryer KM. Fern phylogeny inferred from 400 leptosporangiate species and three plastid genes. Taxon. 2007;56:1037-50. 37. Bierhorst DW. Morphology of vascular plants. New York: Macmillan; 1971. 38. Schuettpelz E, Schneider H, Huiet L, Windham MD, Pryer KM. A molecular phylogeny of the fern family Pteridaceae: Assessing overall relationships and the affinities of previously unsampled genera. Mol Phylogenet Evol. 2007;44:1172-85. 39. Rothfels CJ, Sundue MA, Kuo L-Y, Larsson A, Kato M, Schuettpelz E, et al. A revised family-level classification for eupolypod II ferns (Polypodiidae: Polypodiales). Taxon. 2012;61:515-33. 40. Mickel JT. The classification and phylogenetic position of the Dennstadtieaceae. In Jeremy AC, Crabbe JA, Thomas BA, editors. The phylogeny and classification of the ferns. London: Academic Press for The Linnean Society of London. 1973. p. 135-44. 41. Taylor TN, Taylor EL, Krings M. Paleobotany: The biology and evolution of fossil plants. 2nd ed. San Diego: Academic Press. 2009. p. 1141 42 Scythe. https://github.com/ucdavis-bioinformatics/scythe. Accessed 5 July 2017.

43. Cox MP, Peterson DA, Biggs PJ. SolexaQA: At-a-glance quality assessment of Illumina second-generation sequencing data. BMC Bioinformatics. 2010. doi:10.1186/1471-2105-11-485. 44. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nat Biotechnol. 2011;29:644-U130. 45. van Dongen S. A cluster algorithm for graphs. Technical Report INS-R0010, National Research Institute for Mathematics and Computer Science in the Netherlands. 2000. http://micans.org/mcl/index.html?sec_thesisetc. Accessed 5 July 2017. 46. Hejnol A, Obst M, Stamatakis A, Ott M, Rouse GW, Edgecombe GD, et al. Assessing the root of bilaterian animals with scalable phylogenomic methods. P Roy Soc B-Biol Sci. 2009;276:4261-70. 47. Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Blood PD, Bowden J, et al. De novo transcript sequence reconstruction from RNA-Seq: reference generation and analysis with Trinity. Nat Protoc. 2013; doi:10.1038/nprot.2013.084. 48. Katoh K, Standley DM. MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Mol Biol Evol. 2013;30:772-80. 49. Talavera G, Castresana J. Improvement of phylogenies after removing divergent and ambiguously aligned blocks from protein sequence alignments. Syst Biol. 2007;56:564-77. 50. Stamatakis A. RAxML: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics. 2014;30:1312-3. 51. Dunn CW, Howison M, Zapata F. Agalma: an automated phylogenomics workflow. BMC Bioinformatics. 2013;14:330. 52. Simao FA, Waterhouse RM, Ioannidis P, Kriventseva EV, Zdobnov EM. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics. 2015;31:3210-2. 53. BUSCO results. https://figshare.com/s/bf999173d04b4c311d46. Accessed 5 July 2017

54. Mirarab S, Reaz R, Bayzid MS, Zimmermann T, Swenson MS, Warnow T. ASTRAL: genome-scale coalescent-based species tree estimation. Bioinformatics. 2014;30:I541-8. 55. Seo TK. Calculating bootstrap probabilities of phylogeny using multilocus sequence data. Mol Biol Evol. 2008;25:960-71. 56. ProtienModelselection.pl. https://github.com/stamatak/standard-raxml/. Accessed 5 July 2017. 57. Rößler R, Galtier J. First Grammatopteris tree ferns from the Southern Hemisphere new insights in the evolution of the Osmundaceae from the Permian of Brazil. Rev Palaeobot Palynol. 2002;121:205-30. 58. dos Reis M, Yang Z. Approximate Likelihood Calculation on a Phylogeny for Bayesian Estimation of Divergence Times. Mol Biol Evol. 2011;28:2161-72. 59. Maddison WP, Maddison DR. Mesquite: a modular system for evolutionary analysis. 2011. http://mesquiteproject.org. Accessed 5 July 2017. 60. Lewis PO, Olmstead R. A Likelihood Approach to Estimating Phylogeny from Discrete Morphological Character Data. Syst Biol. 2001;50:913-925. 61. Shen H, Jin D, Shu J, Zhou XL, Lei M, Wei R, et al. Supporting data for "Large scale phylogenomic analysis resolves a backbone phylogeny in ferns" GigaScience Database. 2017. http://dx.doi.org/10.5524/100353

Figure 1. Topologies (a-f) adapted from published results [5, 12-14, 26, 34]. Branches with support < 75% were shown using dotted lines; and taxa which differ in their phylogeny locations were shown in different colors.

Figure 2. A working flow diagram showing the major processes of data production and analysis in this study. Three major processes are de novo transcriptome assembly, one-to-one orthologs prediction, and phylogenetic analysis. The rectangles represent the main results and the ellipses represent the main methods and analysis.

Figure 3. Phylogeny of ferns reconstructed by coalescent-based method using nucleotide sequence with divergence times calculated. Support values for the main phylogeny (a) calculated from Matrix 1/Matrix 2 are listed as percentages; * indicates 100%/100%. Representative leave(s), sporangium and the corresponding lineage are labeled with a same number. Simplified topology (b) shows the main linages as in Figure 1. Species in phylogeny (a) and the corresponding lineage in topology (b) are shown in a same color.

Figure 4. Reconstruction of the evolutionary history of sporangial annulus in ferns. Sampled species with seven types of sporangial annulus are shown in different colours. For each ancient node, percentage of character state of sporangial annulus is shown. Table 1. Sequencing and assembly information of the transcriptome data. The number of ortholog genes used in Matrix 1 and Matrix 2 were shown. ID Species Clean data (G) Total reads (clean) Q30% Number of contigs N50 (bp) Mean (bp) Genes in Matrix 1 Genes in Matrix 2 RS1 Pronephrium simplex 4.7 38045864 91.24 151319 887 581.07 2,168 1,254 RS10 Antrophyum callifolium 4.0 32745384 91.76 64107 1819 998.73 2,226 1,305 RS101 Oleandra musifolia 4.5 36487068 91.45 37075 1493 919.3 2,093 1,248 RS103 Woodsia polystichoides 3.9 31465870 90.91 47812 1348 811.3 2,287 1,310 RS107 Equisetum diffusum 4.4 35693238 90.21 88932 1154 655.64 1,811 1,254 RS108 Oreogrammitis dorsipila 4.6 37037324 90.57 266540 591 485.1 2,141 1,273 RS11 Vandenboschia striata 4.8 38639790 90.3 261724 460 422.76 1,959 1,276 RS111 Pleurosoriopsis makinoi 4.8 38983796 90.13 98187 1145 632.29 2,182 1,277 RS112 Azolla pinnata subsp. asiatica 4.4 35735206 90.57 78295 1348 777.92 1,418 839 RS114 Taenitis blechnoides 4.1 32898682 90.98 70495 1262 711.3 2,186 1,278 RS115 Gymnogrammitis dareiformis 3.9 31630988 89.81 119483 569 449.38 1,996 1,220 RS116 Schizaea dichotoma 4.5 36668734 89.6 67422 1350 826.92 2,035 1,285

RS119 Botrychium japonicum 4.8 38603000 90.28 85236 1477 846.97 1,866 1,283 RS122 Goniophlebium niponicum 4.8 38786214 90.82 54152 1663 951.92 2,279 1,300 RS123 Arthropteris palisotii 4.4 35646740 91 50700 1454 891.67 2,286 1,311 RS124 Matteuccia struthiopteris 4.2 34080998 90.44 57514 1345 776.52 2,290 1,313 RS127 Salvinia natans 4.2 33780056 91.17 79393 1379 767.14 1,905 1,173 RS128 Woodwardia prolifera 5.1 40967322 91.63 69931 1557 859.72 2,328 1,328 RS14 Diplazium viridescens 4.0 32320416 90.46 88236 1434 780.87 2,269 1,310 RS16 Bolbitis appendiculata 4.7 37503336 91.66 201426 802 556.39 2,226 1,288 RS17 Dryopteris pseudocaenopteris 4.1 33136196 91.23 102751 723 514.92 2,236 1,298 RS18 Dicranopteris pedata 4.2 33942120 92.04 74011 1193 684.09 2,031 1,304 RS19 Haplopteris amboinensis 4.2 42772168 94.17 47603 1713 1041.8 2,249 1,307 RS21 Psilotum nudum 8.5 85199034 93.6 66212 1739 927.19 1,741 1,223 RS24 Cyclopeltis crenata 4.6 37158058 91.5 29668 600 491.82 2,146 1,279 RS25 Asplenium formosae 4.6 46629754 93.5 73318 1722 989.84 2,273 1,312 RS27 Lomariopsis spectabilis 4.1 33233594 91.77 98030 1466 750.42 2,225 1,304 RS28 Cheiropleuria bicuspis 5.1 41617294 91.35 99411 1435 832.82 2,022 1,295 RS31 Plagiogyria japonica 5.7 46472760 91.92 89532 1258 733.9 2,036 1,222 RS34 Alsophila podophylla 4.9 48768608 93.43 66254 1580 904.62 2,195 1,289 RS35 Histiopteris incisa 4.3 43115390 93.81 61231 1749 985.03 2,319 1,316 RS36 Pteris vittata 4.1 41212858 94.37 76666 1868 1021.13 2,296 1,312 RS37 Cibotium barometz 4.1 33263550 91.92 85555 1612 891.87 1,790 1,099 RS38 Osmunda japonica 4.1 33485274 92.05 58612 1730 901.28 1,732 1,159 RS39 Loxogramme chinensis 3.9 31392952 92.16 84796 1065 651.88 2,240 1,305

RS4 Microlepia hookeriana 4.0 40561422 94.49 95951 1610 874.06 2,262 1,301 RS41 Pteridium aquilinum 4.6 46157134 93.51 55615 1742 960.37 2,321 1,316 RS42 Hypolepis punctata 4.4 43828154 93.56 59717 1371 833.68 2,277 1,308 RS43 Dicksonia antarctica 3.9 31210608 91.69 56494 1533 902.96 2,045 1,213 RS45 Rhachidosorus mesosorus 4.4 35348994 91.98 80069 1541 835.92 2,300 1,315 RS46 Drynaria bonii 4.5 36017548 92.02 68132 1077 643.93 2,176 1,279 RS47 Platycerium bifurcatum 4.1 33209740 91.62 40456 1097 694.56 2,148 1,283 RS48 Angiopteris fokiensis 4.4 35120302 91.12 57637 1629 932.57 1,917 1,306 RS5 Diplaziopsis brunoniana 4.3 34698846 91.35 70184 822 541.31 2,040 1,234 RS50 Dennstaedtia pilosella 4.5 45618446 93.63 84813 1582 831.56 2,308 1,313 RS51 Monachosorum henryi 4.1 41658504 93.42 87832 1465 803.17 2,255 1,288 RS52 Acystopteris japonica 5.5 44662146 91.15 57118 1507 873.59 1,222 677 RS53 Monachosorum maximowiczii 4.8 48497004 93.58 101448 1817 899.54 2,257 1,294 RS54 Dennstaedtia scabra 5.1 51360716 93.47 92158 1565 845.44 1,818 1,056 RS56 Arachniodes nigrospinosa 5.1 50929362 94.47 57168 1623 916.1 2,332 1,319 RS69 Cheilanthes chusana 5.2 51851066 94.18 49449 1727 1012.63 2,317 1,324 RS7 Elaphoglossum mcclurei 4.1 32800248 92.31 57330 1398 846.79 2,267 1,299 RS70 Lomagramma matthewii 4.4 35218876 91.21 65170 1748 947.18 2,258 1,307 RS71 Osmolindsaea odorata 4.6 46808646 94.13 113778 1521 845.96 2,257 1,312 RS72 Aleuritopteris chrysophylla 4.8 47955674 94.18 61637 1669 929.63 2,307 1,322 RS77 Marsilea quadrifolia 4.3 34724432 91.76 65227 1607 930.31 2,188 1,299 RS8 Humata repens 4.5 36606746 91.17 68932 1267 690.35 2,264 1,315 RS81 Tectaria subpedata 4.2 42539482 94.43 57384 1326 797.83 2,128 1,242

RS84 Ophioglossum vulgatum 4.4 35637330 91.77 71821 1226 741.62 1,631 1,179 RS85 Nephrolepis cordifolia 5.0 40063236 90.81 55207 1530 842.63 2,302 1,319 RS86 Microlepia platyphylla 4.6 46324294 94 74956 1763 945.87 2,267 1,295 RS88 Lygodium flexuosum 4.2 34098316 91.44 66751 1514 867.82 2,064 1,296 RS89 Hypodematium crenatum 4.1 32711798 91.58 52813 1416 852.57 2,298 1,319 RS90 Acrostichum aureum 5.4 43422574 90.69 46189 1729 1043.2 2,303 1,319 RS91 Adiantum caudatum 5.1 51062204 94.23 51145 1575 950.49 2,323 1,327 RS92 Parahemionitis cordata 4.1 33309450 91.72 47508 1456 894.42 2,306 1,317 RS93 Microlepia speluncae 4.4 44124842 94.55 94980 1720 917.59 2,292 1,308 RS97 Stenochlaena palustris 4.7 37887642 91.81 58416 1655 945.83 2,300 1,316 RS98 Ceratopteris thalictroides 3.9 31741082.0 91.4 74728 1610 912.26 2,231 1,296

Table 2. Inconsistent topologies using different methods and sequences. Sit Coalescent-based method Concatenation-based method e nucleotide amino-acid nucleotide amino-acid A (Anfo,(Pnu,(Ovu,Bj (Anfo,(Pnu,(Ovu,Bj ((Pnu,(Ovu,Bja)),(Anfo ((Pnu,(Ovu,Bja)),(Anfo a))) a))),#)),#)) B (Cbi,(Dpe,Vst)) (Cbi,(Dpe,Vst)) (Cbi,(Dpe,Vst)) ((Dpe,Vst),(Cbi,#)) C (Asfo,(Aja,(Dbr,#))) (Asfo,(Aja,(Dbr,#))) (Asfo,(Aja,(Dbr,#))) (Asfo,((Aja,Dbr),#)) D (Dvi,(Mst,(Spa,Wpr) ((Dvi,Mst),(Spa,Wpr) (Dvi,(Mst,(Spa,Wpr))) (Dvi,(Mst,(Spa,Wpr))) )) ) E (Bap,(Emc,Lma)) (Emc,(Bap,Lma)) (Bap,( Emc,Lma)) (Emc,(Bap,Lma)) F (Nco,((Tsu,Apa),#)) (Nco,(Tsu,(Apa,#))) (Nco,((Tsu,Apa),#)) (Nco,((Tsu,Apa),#)) (A) Anfo: Angiopteris fokiensis, Pnu: Psilotum nudum, Ovu: Ophioglossum vulgatum, Bja: Botrychium japonicum; (B) Cbi: Cheiropleuria bicuspis, Dpe: Dicranopteris pedata, Vst: Vandenboschia striata; (C) Asfo: Asplenium formosae, Aja: Acystopteris japonica, Dbr: Diplaziopsis brunoniana; (D) Dvi: Diplazium viridescens, Mst: Matteuccia struthiopteris, Spa: Stenochlaena palustris, Wpr: Woodwardia prolifera; (E) Bap: Bolbitis appendiculata, Emc: Elaphoglossum mcclurei, Lma: Lomagramma matthewii; (F) Nco: Nephrolepis cordifolia, Tsu: Tectaria subpedata, Apa: Arthropteris palisotii. # indicates other sampled species within this lineage. Topologies consistent with the one yielded from coalescent-based method and nucleotide sequences are shown in bold.